1. Field of the Invention
The present invention relates generally to character recognition using a camera, and more particularly, to an apparatus and method for optimizing user views by adjusting a size of characters in an image received by means of a camera.
2. Description of the Related Art
Camera-based Optical Character Recognition (OCR) may be classified into Snapshot OCR (OCR based on a view of an image to be photographed) and Preview OCR (OCR based on a preview image or viewfinder images). In SnapShot OCR, an image is acquired according to the same method as a normal method for acquiring an image for a photograph. The image acquisition process includes, in turn, photographing desired characters by means of a camera, performing a character recognition process on the photographed characters, and outputting the recognition results to an output unit.
By contrast, in Preview OCR, images are input at a preset rate of Frames. Per Second (FPS). Therefore, in Preview OCR, character recognition is possible in real time. Generally, an image acquisition process of Preview OCR includes scanning characters by a user by means of a camera, and continuously performing a recognition operation on an image received through the camera using a recognizes in the camera.
The largest difference between Snapshot OCR and Preview OCR consists in the resolution of an image to be recognized. While Preview OCR recognizes a small image of Quater Video Graphics Array (QVGA; 320×240) for real-time recognition, Snapshot OCR recognizes an image of 1 Mega pixels or more, thus having a longer recognition time compared with Preview OCR. Due to the use of high-resolution images, Snapshot OCR is far superior to Preview OCR in character recognition capability. Recently, as the performance of cameras improves and accurate recognition becomes increasingly important. Snapshot OCR has become widely used than Preview OCR.
It is possible for Snapshot OCR to recognize all characters in an input image, or characters corresponding to a particular region among all the characters in the input image. Hence, a user may perform full recognition on the image displayed on a Liquid Crystal Display (LCD) screen of the camera, or perform partial recognition on a particular region designated.
FIG. 1 illustrates a conventional process of recognizing characters in an image.
In step 101, a mobile terminal receives an image with characters using a camera module. In step 103, the mobile terminal may acquire a focused image from the received image using an Automatic Focusing (Auto-Focus) technique. In step 105, the mobile terminal stores the acquired focused image in a separate storage. In step 107, the mobile terminal recognizes characters in the stored image. The mobile terminal may recognize all the characters in the image or only characters in a region of the image selected by the user. In step 109, the mobile terminal displays the recognition results on its display.
Since, unlike the conventional name-card recognition, Snapshot OCR has no limitation on a recognizable distance desired by the user, characters in the input image may vary in size. For example, when an image, in which very large characters and very small characters are included in various ways, is resized on an LCD, the sizes of the displayed characters are also appear very small and very large, proportionally to the actual image. The excessively large or small size of the characters displayed on the LCD may reduce recognition efficiency or cause recognition failure.